Chronicles-OCR - Mature Script Classification: leaderboard

Chronicles-OCR (2026) tests vision-language models on historical Chinese writing across the Seven Chinese Scripts, 2,800 expert-annotated images from Oracle Bone to Cursive. This board scores naming which of the seven scripts a Clerical, Regular, Running or Cursive image shows, as accuracy in percent averaged over the four scripts; higher is better.

Metric: Accuracy (%) of classifying the image into one of the Seven Chinese Scripts, on images of the mature scripts (Clerical, Regular, Running and Cursive), the paper Average column over the per-script results; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 26 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Kimi K2.5Non-reasoning77#266
2Seed 2.0 ProNon-reasoning76.1#222
3Gemini 3.1 Pro (Preview)73.1#99
4Qwen 3.5 397B A17BNon-reasoning72.2#264
5Seed 1.871.1#252
6Qwen 3.5 35B A3BNon-reasoning68.1#492
7Claude Opus 4.7Thinking66.8#85
8Qwen 3 VL 235B A22B Instruct64.9#532
9MiMo-V2-Omni62.3#465
10GPT-562.1#178
11Qwen 3 VL 8B Instruct60.9#1029
12Qwen 3 VL 235B A22B (Thinking)60.4#459
13InternVL3.5-241B-A28B58.1#733
14Qwen 2.5 VL 72B Instruct57.2#841
15Gemma 4 31B (IT)57.1#350

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=chronicles-ocr-mature-script-classification · How It Works · Data refreshed daily, snapshot 2026-11-21.